rethinking geocoding in utah

17
Rethinking Geocoding Bert Granberg AGRC

Upload: bert-granberg

Post on 16-Apr-2017

689 views

Category:

Technology


1 download

TRANSCRIPT

Page 1: Rethinking Geocoding In Utah

Rethinking Geocoding

Bert GranbergAGRC

Page 2: Rethinking Geocoding In Utah

Paradigm shift

• 2000:– Geocoding Addresses was cool...

• “Wow, it found it!”

• 2010:– Accurately Locating Addresses is critical

• “Every address must match!”

Page 3: Rethinking Geocoding In Utah
Page 4: Rethinking Geocoding In Utah

Basic GIS Locator & Process• Input: Structured Address(es)• Output: Match Candidates and Scores

• Familiar but soooo dated

• Many shortcomings:

– Input Address Data• Un-standardized input addresses• Incorrect zones (zip codes in particular)

– Imperfect Address Source Data• Imperfect zone information• Prefix direction used inconsistently…USPS standard in conflict

with local address vernacular

Page 5: Rethinking Geocoding In Utah

Shortcomings (cont.)– Process Too Simple

• Single address source dataset (centerlines)• Single search strategy (indexing) ….composite and custom locators allow

more complex logic• Focus on HN + StreetName addresses, what about route/milepost or non-

addressed locations• User expectation for more flexible, simple input interface (google)???

– Output• QCing of results done rarely• Improvements to data and process not easily shared• Historical addresses are next to impossible to find• Range-based, interpolated point may not even be on the property• Address point may not be at preferred location (centroid, frontage, building,

doorstep, driveway, gate)

– Overall All• Different users get imperfect, inconsistent results

Page 6: Rethinking Geocoding In Utah

Geocoding at AGRC

• “Local data is better data”• “Open, public domain data makes sense”

– Cost– Control– Currency

Page 7: Rethinking Geocoding In Utah

Utah Registered Voter Geocoding Rate By County, May 2010

Overall Match Rate, Weighted By Population = 95%

0%

80%

60%

40%

20%

100%

Page 8: Rethinking Geocoding In Utah

Geocoding at AGRC (cont.)

• Web service based• Address, Zone (2 parameters)

• 3 levels of aliasing (composite address locator)

• 95% match statewide (range from 47-98% by county) VISTA database

• Can use zip or incorporated city for “zone”

Page 9: Rethinking Geocoding In Utah

GCWeb

Service

Page 10: Rethinking Geocoding In Utah

Commitments Made:

• 911 Committee Goals– 95% Address Match Rate by June 2011– Statewide Address Points June 2012

• VISTA, Elections Management– GIS Precinct Assignment and Counts

• Broadband Project funded Address Point project– 1-2 year time frame for building sustainable

processes– Collect Once application

Page 11: Rethinking Geocoding In Utah

In progress• Developing Address Point-based location services

– Parcel to Address Polygon to Address Point 'Starting Point' script– Proposed address point data model

• Street Centerline– Incorporate 'address system' as an attribute for potential zone use– Prefix and suffix direction issues

• Developing advanced input address handling logic

• Developing advanced address location logic– Different classes of addresses get handed to different location service

• Functionality to gather improvement data from batch and web app results and pass along to data stewards

• Magic search box beginning to take shape

• New generation, distributed data sharing portal, planning phases

Page 12: Rethinking Geocoding In Utah

Address Starting Points from parcels

http://gis.utah.gov/code-visual-basic/vba-experiment-building-address-points-from-parcels

Page 13: Rethinking Geocoding In Utah

Magic Zoom

Page 14: Rethinking Geocoding In Utah

Input Address Classes• _HN _AlpName (123 Main)• _HN _Pre _AlpName (123 S Main)• _HN _Pre _ AlpName _ Type + Zone (123 S Main St)• _HN _AlpName _ Type + Zone (123 Main St)• _HN _Pre _NumName _Suf + Zone (123 S 300 E)• _Name _IntSep _Name + Zone (Main St and 600 S)• _Milepost + _Route (123.456 NB I-15)• _GeogName (Delicate Arch)• _GeogName + Zone (Riverwoods Conference Center, Logan)

Where Zone = {Zip, City, AddressSystem}

Page 15: Rethinking Geocoding In Utah

Advanced Match Logic SketchWith Input Address:

• Is Geographic Place?

• Is Milepost & Route?

• Is Intersection?

• Is Numeric Street Type?

• Has Prefix Direction?– Lookup and use Address System Zone from supplied zip/city– Use Prefix enabled locators

• Has No Prefix Direction, Prefix Not Critical for Streetname Within Zone?

– Lookup and use Address System Zone from supplied zip/city– Use Prefix disabled locators

• Has No Prefix Direction, Prefix Is Critical For StreetName Within Zone?

– Use supplied Zip/City Zone– Use Prefix disabled locators– Guess best match? – Return match note that prefix should be used

Page 16: Rethinking Geocoding In Utah

Other, future considerations

• Interim study on Address Quality in government databases (every address locatable?)

• Centralized Repository & Services• Quality Assessment feedback loop• Utah Addressing Board???• Non-Public Sector Services

– Commercial ??– Open APIs ??

Page 17: Rethinking Geocoding In Utah

Comments, Questions?

• Bert Granberg– [email protected]– @BertAGRC